Exploring Plain Vision Transformer Backbones for Object Detection
时间: 2024-04-24 11:24:49 浏览: 166
回答: 本文探索了将普通的Vision Transformer (ViT)作为目标检测的骨干网络。通过对ViT架构进行微调,而无需重新设计分层骨干进行预训练,我们的普通骨干检测器可以取得竞争性的结果。研究发现,只需从单尺度特征图构建简单的特征金字塔(无需常见的FPN设计),并使用窗口注意(无需移动)辅助少量的跨窗口传播块即可。通过使用预先训练的纯ViT主干作为Masked Autoencoders (MAE),我们的检测器ViTDet可以与之前基于分层骨干的领先方法竞争,在COCO数据集上达到61.3 APbox的性能。我们希望这项研究能够引起对普通骨干检测器的关注。\[1\]\[2\]\[3\]
#### 引用[.reference_title]
- *1* [论文阅读-ViTDet:Exploring Plain Vision Transformer Backbones for Object Detection](https://blog.csdn.net/qq_37662375/article/details/126675811)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v91^insert_down28v1,239^v3^insert_chatgpt"}} ] [.reference_item]
- *2* *3* [ViTDet:Exploring Plain Vision Transformer Backbonesfor Object Detection(arXiv 2022)](https://blog.csdn.net/qq_54828577/article/details/127262932)[target="_blank" data-report-click={"spm":"1018.2226.3001.9630","extra":{"utm_source":"vip_chatgpt_common_search_pc_result","utm_medium":"distribute.pc_search_result.none-task-cask-2~all~insert_cask~default-1-null.142^v91^insert_down28v1,239^v3^insert_chatgpt"}} ] [.reference_item]
[ .reference_list ]
阅读全文